A Data-Driven Intelligent Management Scheme for Digital Industrial Aquaculture based on Multi-object Deep Neural...

Area of Science:

• Aquaculture
• Artificial Intelligence
• Data Science

Background:

• Traditional aquaculture relies on manual observation, limiting comprehensive monitoring of fish and water quality.
• The industry is transitioning towards intelligent, data-driven industrial models.

Purpose of the Study:

• To propose a data-driven intelligent management scheme for digital industrial aquaculture.
• To enhance fish state and environmental state management using advanced neural networks.

Main Methods:

• Developed a multi-object deep neural network (Mo-DIA) for intelligent aquaculture management.
• Utilized a double hidden layer BP neural network for predicting fish weight, oxygen consumption, and feeding amounts.
• Employed an LSTM neural network to model temporal correlations in water quality data for predicting eight attributes.

Main Results:

• The Mo-DIA scheme effectively manages fish and environmental states in digital aquaculture.
• The BP neural network accurately predicted key fish metabolic and growth indicators.
• The LSTM model demonstrated high accuracy in forecasting multiple water quality parameters.

Conclusions:

• The proposed Mo-DIA offers an effective and accurate solution for intelligent aquaculture management.
• This data-driven approach addresses the limitations of manual observation in modern aquaculture.
• The system supports optimized fish farming through precise prediction of fish and environmental conditions.